The FDIC’s brokered deposits framework was created in 1989 in response to the savings and loans crisis of the 1980s. Section 29 of the Federal Deposit Insurance Act aimed to protect the Deposit Insurance Fund by restricting less-than-well-capitalized banks’ ability to accept brokered deposits at above-market rates.[1]
Over time, the FDIC interpreted the definition of deposit broker expansively, leading to deposits being considered brokered that did not bear the same characteristics of the “hot money” that was the target of Section 29. And the consequences of this expansive interpretation extend further by penalizing banks for holding “brokered deposits” even if the bank is well capitalized.[2]
In 2020, the FDIC adopted revised rules to address the scope of the regulatory definition of brokered deposits given technological innovations and the modern provision of financial services. Among the types of deposits addressed by the FDIC’s 2020 rule are so-called “sweep” deposits. These are programs where broker-dealers place customers’ investment account cash balances with insured depository institutions (banks) for safekeeping. These banks may or may not be affiliated with the broker-dealer, and these arrangements provide investors a safe place for their investment account’s free cash balance. The FDIC’s 2020 rule clarified that deposits from sweep programs are not brokered if less than 25 percent of customer total assets were placed in banks, reasoning that such broker-dealers do not have the “primary purpose” of placing funds at banks.[3]
The FDIC’s 2024 proposal would substantially narrow the availability of this exclusion. For example, under the proposed “fee prong,” any additional third party receiving compensation for sweep program services would be classified as a deposit broker, unless the FDIC approves an exception through an application process. The proposal would also make two other significant changes to the primary purpose exclusion for sweeps: it would reduce the asset threshold from 25 percent to 10 percent and base calculations on “assets under management,” instead of “assets under administration.” This shift could restrict the exclusion to only actively managed accounts, excluding self-directed investment accounts where investors control their investments.
Furthermore, the FDIC has solicited comments on more stringent alternatives. These include the complete elimination of the designated business exception for broker-dealer sweeps or limiting it exclusively to deposits placed at affiliated banks. These restrictions would apply across all business lines, thus making these alternatives even more stringent. Under either alternative scenario, broker-dealer sweep deposits to unaffiliated banks would be universally classified as brokered unless specifically exempted through an FDIC-approved primary purpose exception application.
This note examines the empirical relationships observed during the March 2023 stress event between bank risk and two deposit categories: brokered deposits and non-brokered sweep deposits. The note then evaluates the FDIC’s proposed regulatory changes in light of these findings. For the purposes of this analysis, “brokered deposits” and “sweep deposits” refer to deposits classified as such on the call report. Any regulatory reclassification of these deposit programs would raise costs for banks and broker-dealers and, in turn, for investors. Furthermore, such changes could also limit the availability of these sweep programs and the benefits they provide to investors.
Our analysis significantly undermines the rationale for the FDIC’s proposed rule. First, the FDIC cites the substantial decline in reported brokered deposits—from $996 billion to $652 billion between March and June 2021—as evidence of potential underreporting. Our analysis demonstrates that 10 institutions account for approximately 90 percent of this decline. Of these, the two largest contributors—accounting for 80 percent of the decline—received a primary purpose exception for broker-dealer sweep arrangements in the second quarter of 2021. Thus, the reclassification of sweep deposits to non-brokered status explains the majority of the observed decline in brokered deposits following the implementation of the 2021 final rule.
Second, our empirical analysis shows that non-brokered sweep deposits exhibit no statistically significant correlation with bank risk after accounting for bank-specific characteristics, indicating fundamentally different risk characteristics from traditional brokered deposits. In contrast, and consistent with academic research, there is a positive correlation between reliance on brokered deposits and bank risk, particularly for banks below $100 billion in assets.
Notably, the FDIC’s proposal contains no contrary analysis in support of its requirements, relying instead on anecdotal evidence. Our systematic analysis suggests that sweep deposits and brokered deposits exhibit distinctly different risk characteristics. This empirical evidence indicates that maintaining separate regulatory treatment for these deposit categories may be more appropriate than the proposed consolidation under the 2024 proposal.
The Temporary Drop in Brokered Deposits Post-2021 Final Rule
The 2024 brokered deposits proposal seeks to reverse the changes to the definition of brokered deposits made by the 2021 final rule, which had provided new exclusions from the “deposit broker” definition. The FDIC’s concern appears to stem from the possibility that these reclassified deposits may retain characteristics and risks traditionally associated with brokered deposits, despite their modified regulatory treatment. Of particular significance, the 2021 final rule established a 25 percent threshold, whereby institutions could place up to a quarter of their customer funds on deposit without triggering deposit broker classification. This provision notably expanded banks’ capacity to accommodate asset managers’ sweep account deposits without incurring the regulatory implications associated with brokered deposit classification.
The goal of the 2024 proposal is to reverse some of these modifications, citing significant shifts in reported brokered deposit levels as evidence of underreporting of brokered deposits and its consequences for the Deposit Insurance Fund. Specifically, banks’ call reports show that total brokered deposits declined from $996 billion to $652 billion—a reduction of approximately $350 billion—between March 31, 2021, and June 30, 2021 (illustrated by the solid line in Figure 1). This substantial decline coincides with the implementation period of the 2021 final rule.
Table 1 shows the 10 banks that reported the largest declines in brokered deposits during the period of March 31, 2021 to June 30, 2021. These institutions collectively account for approximately 90 percent of the aggregate decline in brokered deposits during this interval, with the two banks at the top of the list representing nearly 80 percent of the total reduction. Data provided by the FDIC indicate that this substantial decline in brokered deposits is attributable to the institutions’ eligibility for a primary purpose exception under the 2021 final rule, specifically for their broker-dealer sweep arrangements. Notices were filed by these institutions in April and June 2021.
Around the same time the 2021 final rule became effective, the FFIEC also adopted changes to the call report to collect more information on sweep deposits, specifically because those deposits would no longer be reported as brokered.[4] This data suggests that the observed decline in brokered deposits in the second quarter of 2021 may be partially attributable to the reclassification of these sweep deposits as non-brokered. As shown in Table 1, a significant proportion of the institutions contributing to the decline in brokered deposits concurrently reported substantial amounts of sweep deposits in the new line items on the call report that took effect as of the Sept. 30, 2021 report date.
The dashed line in Figure 1 represents a counterfactual analysis that estimates brokered deposits level absent the regulatory reclassification of sweep deposits to non-brokered status. This counterfactual analysis demonstrates that the reclassification of sweep deposits represents the main explanation for the observed decline in brokered deposits following the implementation of the 2021 final rule.
The 2024 proposal points to the fact that brokered deposits declined under the 2021 rule as support for the proposed changes. However, the decline in brokered deposits on its own does not provide any evidence of the need for changes and warrants more systematic empirical investigation.
Empirical Analysis
This section examines the relationship between banks’ reliance on brokered deposits and sweep deposits and their stock returns (as a proxy for bank risk) during the significant stress event of the March 2023 mid-sized bank crisis.
Our primary analysis focuses on bank equity returns during March 2023, a period characterized by significant stress in mid-sized banking institutions following Silicon Valley Bank’s failure. SVB’s collapse was idiosyncratic but led to market focus on other institutions that shared certain similar characteristics and many of those institutions experienced considerable market stress.
Panel A of Figure 2 illustrates the negative relationship between bank holding companies’ equity returns during March 1-13, 2023, and their ratio of brokered deposits to total domestic deposits as of Dec. 31, 2022.[5] As shown in Panel B, however, the correlation between stock returns and reliance on brokered deposits varies significantly by bank size. The correlation is only negative for banks with less than $100 billion in assets, particularly those between $10 billion and $100 billion. Our regression results below demonstrate that the negative correlation exists independently of the adverse press coverage smaller banks received during the March 2023 mid-sized bank crisis. Conversely, banks above $100 billion in size exhibit a positive correlation between stock returns and reliance on brokered deposits.[6]
By contrast, Figure 3 demonstrates that sweep deposits not classified as brokered deposits showed virtually no correlation with bank stock performance during the March 2023 stress period for banks in any size category. This finding suggests that markets viewed these sweep arrangements differently from traditional brokered deposits, likely due to their distinct operational characteristics and customer relationships. The lack of correlation persists across different bank size categories (not shown).
To formally test these relationships, we estimate the following regression specification:
The dependent variable, ri, represents cumulative excess returns for bank holding company i, while our key explanatory variables Zi,1 and Zi,2 denote the ratios of brokered deposits and total sweep deposits that are not brokered deposits to total domestic deposits, respectively. The vector Xi includes all variables that are important to include in the regression to analyze the relationship between brokered and sweep deposits and bank risk.
We control for bank size by segmenting the sample into three categories: banks with less than $10 billion in assets, those between $10-$100 billion, and banks exceeding $100 billion in total assets. The amount of high-quality liquid assets inclusive of unrealized losses (measured as a percentage of total assets) captures the bank’s ability to meet short-term obligations and withstand funding shocks. We included the amount of unrealized losses on all securities since this factor drove much of the negative press coverage around banks during the March 2023 stress period.
We also include uninsured deposits (measured as a percentage of total assets). While this publicly available metric gained attention during the spring 2023 banking stress, it’s important to note that it may not accurately reflect funding stability. Different business models, particularly those of banks where deposits are primarily operational in nature, may maintain higher levels of uninsured deposits without necessarily indicating increased risk. Other types of uninsured deposits such as collateralized and affiliate deposits also have features that are likely to make them more stable than other types of uninsured deposits. We include this measure because it was a key metric that market participants focused on during that period.
Return on assets serves as a proxy for operational efficiency and earnings capacity. Finally, the Tier 1 leverage ratio controls for capital adequacy and loss-absorption capacity.[7]
Table 2 presents our regression estimates. The first two columns examine the relationship between brokered deposits and equity returns. Column (1) reports results for the placebo period (Jan. 1 – Feb. 28, 2023), while Column (2) presents findings for the period of market stress (March 1-13, 2023). The coefficient estimates indicate that prior to the March 2023 banking crisis, the reliance on brokered deposits exhibited a positive association with stock returns. However, this relationship reversed significantly during the crisis period, with the coefficient becoming negative and both economically and statistically significant.
Specifically, a 10-percentage-point increase in the ratio of brokered deposits to total deposits is associated with a nearly 3-percentage-point decline in stock returns during the crisis period, representing approximately a decline from the median stock return to the first quartile stock price decline observed during the March 2023 period. Moreover, the coefficient on sweep deposits not classified as brokered is statistically insignificant, suggesting no discernible relationship with bank risk.
The estimated coefficients of the remaining control variables largely align with our expectations. The coefficient on high-quality liquid assets is positive and statistically significant at the 5 percent level, indicating that banks with stronger liquidity positions experienced less severe declines in stock returns during the period of market stress. Consistent with market focus during this period, the coefficient on uninsured deposits is negative and statistically significant at the 1 percent level, suggesting that banks with higher proportions of uninsured deposits experienced larger declines in their stock returns. Profitability, as measured by return on assets, exhibits a positive and statistically significant relationship with stock returns, implying that more profitable institutions demonstrated greater resilience during the stress period. Notably, the Tier 1 leverage ratio coefficient lacks statistical significance, suggesting that regulatory capital levels did not meaningfully influence market perceptions of bank risk during this period.
The remaining columns in Table 2 expand our analysis by examining how brokered deposits relate to bank risk when segmented by bank size (columns 3 and 4). This segmented analysis confirms the findings illustrated in Figure 2: the positive correlation between reliance on brokered deposits and bank risk is observed only for banks below $100 billion in assets. As in the baseline case, the coefficient on sweep deposits not classified as brokered remains statistically insignificant, suggesting no discernible relationship with bank risk. This distinct behavior of non-brokered sweep deposits during the period of market stress supports the current regulatory treatment that allows most sweep deposit arrangements to be classified separately from brokered deposits.
Columns (5) and (6) in Table 2 further disaggregate sweep deposits into affiliated and non-affiliated components. During the crisis period, the coefficient estimates for both types of sweep deposits are statistically indistinguishable from zero at conventional significance levels. This demonstrates that both affiliated and non-affiliated sweeps did not have a discernible relationship with bank risk.
In summary, the findings reported in Table 2 show a fundamental difference between sweep deposits and brokered deposits, casting doubt on the justification for recent regulatory proposals that would expand the classification of sweep deposits as brokered deposits.
Conclusion
Our empirical analysis of the 2024 brokered deposits proposal provides two important findings regarding the reversal of the narrowing of the types of deposit related activities that are considered to be brokered.
- Based on a counterfactual analysis, we show that the reclassification of sweep deposits from brokered to non-brokered status likely accounts for most of the observed $350 billion decline in reported brokered deposits following the implementation of the 2020 Final Rule.
- Econometric analysis of bank performance during a period of market stress—specifically the March 2023 mini-banking crisis—indicates that sweep deposits exhibit markedly different risk characteristics compared to brokered deposits. While smaller banks with higher concentrations of brokered deposits experienced significant negative excess returns during stress periods, we find no statistically significant relationship between sweep deposit concentrations and bank risk when controlling for relevant bank characteristics among banks of any size.
These empirical findings raise substantive questions about the FDIC’s proposed regulatory changes, which seek to reverse the 2021 final rule’s treatment of brokered deposits based primarily on anecdotal evidence from recent bank failures. Our systematic analysis suggests that maintaining the changes adopted in the 2021 final rule better reflect the underlying risk characteristics of brokered and sweep deposits.
[1] Senate Congressional Record, Proceedings and Debates of the 101st Congress, First Session, 135 Cong. Rec. S4238-01, 1989 WL 191889 (Apr. 19, 1989).
[2] These penalties include (1) higher deposit insurance assessment rates; (2) requirements to maintain additional liquid assets to offset the higher outflow rates assigned to brokered deposits under the liquidity coverage ratio (brokered deposits also are generally assigned lower Available Stable Funding factors under the net stable funding ratio); (3) for GSIBs, higher GSIB surcharges that require GSIBs to hold higher levels of total loss absorbing capacity; (4) for Category III and Category IV institutions, potentially higher liquidity coverage ratio and net stable funding ratio requirements than would otherwise apply; (5) expending additional time and resources on liquidity planning (e.g., incorporating PCA-related downgrade triggers into contingency funding plans); and (6) reduced ability to compete with nonbank competitors that are not required to factor liquidity costs into product offerings that would result in brokered deposits for a bank (e.g., money market mutual funds).
[3] The statute excludes deposits from being considered brokered if the “primary purpose” of the third party “is not the placement of the funds with depository institutions.” 12 USC 1831f(g)(2)(I).
[4] 86 FR 6742, 6761 (noting “institutions will be required to report to the FDIC or on the Call Report certain types of deposits that will not be considered brokered deposits under the final rule” and “[t]he FDIC plans to monitor the data resulting from such reporting and will consider in the future whether modifications to deposit insurance assessment pricing related to certain types of funding concentrations are warranted, consistent with the statutory requirement that the assessments be risk-based); 86 FR 8480, 8484 (describing agencies’ intent “to update the Call Report to obtain data that will assist in better evaluations of funding stability for sweep deposits over time to determine their appropriate treatment under applicable liquidity regulations and to assess the risk factors associated with sweep deposits for determining their deposit insurance assessment implications, if any.”)
[5] The choice of dates follows recent academic literature analyzing the March 2023 mini-banking crisis. For example, Acharya, Viral V and Chauhan, Rahul S and Rajan, Raghuram and Steffen, Sascha, “Liquidity Dependence and the Waxing and Waning of Central Bank Balance Sheets”, National Bureau of Economic Research, Working Paper number 31050, March 2023, available at http://www.nber.org/papers/w31050.
[6] The sample includes 201 publicly traded banks with less than $10 billion in assets, 78 banks between $10-100 billion, and 25 banks exceeding $100 billion in total assets.
[7] Not all banks in our sample report risk-based capital ratios, as some opted into the community bank leverage ratio framework. We also included a measure of commercial real estate loans concentration (measured as a percentage of total assets) to control for sectoral exposure vulnerability—particularly relevant given the significant impact of the COVID-19 pandemic on this asset class. In addition, the nonperforming loans ratio was included to capture asset quality and credit risk. However, these two variables were never statistically significant, so they were dropped from the final specification.
